Download pdf - Studies on Ziegler-Natta catalysts. Part II. Reactions between α- or β-TiCl3 and AlMe3, AlMe2Cl, or AlEt3 at various temperatures

JOURNAL OF POLYMER SCIENCE: PART A-1 VOL. 4, 1917-1926 (1966)

Studies on Ziegler-Natta Catalysts. Part 11. Reactions between (Y- or p-TiCl, and AlMe,, AlMe,Cl, or AlEt, at Various Temperatures

L. A. 11. RODRIGUEZ,* H. M. VAN LOOY,f and J. A. GABANT, Union Carbide European Research Associates, 8. A. , Brussels, Belgium

synopsis

The following reactions, carried out in the absence of solvents, has been studied: a-TiCh + Al(CH3)a at 20°C., p-TiCl3 + Al(CH3)a at 65"C., a-TiCL + Al(CHS)&l at 20 and 65"C., and a-TiCla + Al(C2Ha)a between 30 and 65°C. It appears that a general reaction mechanism, such as discussed in the preceding paper of this series, applies to all these reactions between TIC13 and aluminum alkyls. The differences in overall stoichiometry between some of these systems may be l i k e d to differences in stability of the intermediate T i 4 bonds. In the case of a-Ticla + .Al( CHa)&l, alkybtion is probably accompanied by fixation of the AlCHaC12 on the nonvolatile product.

I. INTRODUCTION

Part I of this series of papers' concerned the reaction between a-TiC13 and Al(CH&, run at 65°C. In this paper results on three related catalyst systems are presented: a-TiCL and A1(CH3)3 at 20"C., p-TiC13 and Al(CH& at 65°C. and a-TiC13 and AlMeC1 at 20 and 65°C. Some results and comments on the reaction between a-TiC13 and A1Et3 are also given.

11. EXPERIMENTAL

The general experimental technique, called method A, has been described earlier.2 In some cases the initial period of the reaction has been followed by the more elaborate method B.' The preparation of most of the reagents has been described.2 AlCl(CH& was obtained by heating A1(CH3)a and AlC13, the latter being in small excess over the stoichiometric quantity, in a sealed tube at 60°C. The product was freed from AlC12CH3 by complex formation with NaC1.

* Present address: Belgium.

t Present address:

UCB Pharmaceutical Division, 68, rue Berkendael, Brussels,

Institut FranCais du P6trole, Rueil-Malmaison, S. et O., France. 1917

1918 L. A. M. RODRIGUEZ, H. M. VAN LOOY, AND J. A. GABANT

III. REACTION BETWEEN (Y-TiCl3 AND AI(CH8)S AT 20°C.

Two reactions were run at 20°C. in the absence of any solvent and fol- The temperature was kept at 20°C. in all steps

The results are summarized in lowed by method A. except the last of the second run (65°C.). Table I.

TABLE I Reaction of a-TiCla + Al(CHa)a, 2OoC., Method A

RB4 RB13

Step 1 Step 1 Step2 Step3 Step4

Step temperature, "C. Surface a-TiCl,, m.2 Ti,,, f .

Clsurf. Partial reaction time,

sec. Total time, see. CL Cl"/Alf ZCl,/ZAlr

Z(CR + CR)/ZAlf (CR + CD)/clv

Clv/TiBurf. ZClv/TiBurf.

20 20 20 243 172 2.37 1.68 3.55 2.52

1140 1200 1200 1140 1200 2400 1.08 0.784 0.181 1.96 1.56 1.12 1.96 1.56 1.46 0.65 0.89 1.3 1.28 1.38 1.44 0.46 0.47 0.11 0.46 0.47 0.57

20

5400 7800 0.091

(1.51) 2 .0

(1.60) 0.05 0.63

(2.5)

65

1200 9000 0.155 0.97 1.41 1.8 1.64 0.09 0.72

The symbols in Table I have been defined in the preceding paper.l The values in parentheses appear to be too high, probably because of experimental error.

I t is clear that the reaction at 20°C. presents the same general features as the reaction at 65°C. : I decrease of the Cl,/Alf and ZC1,/ZAlf ratios, increase of the (CR + CD)/Cl, and Z(CR + CD)/ZAlf ratios. It follows that the general reaction scheme applies to both temperatures. More particularly, it may be seen from the results of experiments RB13 and RB4 that chlorine is removed rapidly even at 20°C. until a Clv/Tisurf. ratio of approximately 0.5 is reached. This is compatible with our conclusion that the first step in the reaction sequence is a fast exchange reaction between Tic13 and Al(CH3)3.

IV. REACTION BETWEEN p-TiCI, AND Al[CH& AT 65°C.

The P-TiCl3 was prepared by reaction between TiCll and Hz in a silent discharge according to the method of Gutmann, Nowotny, and OfnerJ3 in which the gaseous ratio of TiC1, and Hz is adjusted to give TiC13 and not TiCl2. The brown product obtained has a large specific surface (- 1.5.5 m.?/g.). Its general morphology cannot be observed even by electron microscopy. The electron micrographs show only a very fine powder without structure (cf. also Korotkov and Li4).

ZIEGLER-NATTA CATALYSTS. I1 1919

0

(b)

Fig. 1. ( a ) Stereochemical model of the stack which characterizes the &modification ( b ) Top view of a face perpendicular to the c axis, of TiCI, according to Natta et a1.6

showing the structure of the most probable faces parallel to the c axis.

The crystalline structure of 8-TiCL was studied by Natta and co- w o r k e r ~ , ~ ~ ~ and their results were confirmed by Cras' working on better crystallized samples.

P-TiC13 has a fiber-shaped structure. TiCI, units are piled up in stacks parallel to the main axis c of the crystal The titanium atoms situated along the c axis are octahedrically coordinated to six chlorine atoms belonging to the same stack. The stacks are linked only by van der Waals forces acting between chlorine atoms of neighboring rows. It seems reason- able, therefore, that the crystal grows mainly along the c axis using the coordination energy and that the faces parallel to the c axis are more devel- oped than the faces perpendicular to the same axis. Our discussion is based on the assumption that the faces perpendicular to the c axis give a negligible contribution to the total surface.

Now, it may be seen from Figure 1 that for any crystal face parallel to the c axis the distance between the titanium atoms is 2.91 A. along the c axis and 6.27 A. in the direction perpendicular to the axis.

Therefore, whatever the face considered and provided that every titanium atom is bound to three chlorine atoms, the external surface corresponding to one titanium atom may be evaluated at 18.25 A.2


TABLE I1 Reaction of @-Tic18 + Al(CH&, 65OC., Method B

RB24

Step temperature, "C. Surface &TIC&, m.2

Tisurf. Partial reaction time, sec. Total time, sec. C1" CL/Alr ZClV/ZAl, (CR + c D ) / c I v

Clv/Tlsurf. Z(CR + Cn)/ZAlr

ZClv/TiBurr.

Step 1

65 195 1, 77 63 63 0.938 1.19 1.19 1.13 1.34 0.53 0.53

Step 2 Step 3

65

294 357 0.427 1.18 1.19 1.55 1.50 0.24 0.77

65

903 1260 0.525 1.29 1.22 1.00 1.45 0.30 1.07

As in the case of the a-TiCL, the volatile compounds after reaction con- t ain only methane, chlorodimethylaluminum and an excess of trimethylaluminum. Because the values of Cl,/Alr and ~ ( C R + CD)/ZAlf are of the same order in both cases, the same general reaction scheme may be applied. 1

However, in contrast to the a-TiCL + Al(CH& reaction, no significant variation of the ratio Cl,/Alt is observed as a function of time. Neverthe- less it is highly probable in the case of the p-TiC13 too, the reaction takes place by a multi-step sequence in which the first step is the exchange (alkylating) reaction followed by the fixation of trimethylaluminum on the alkylated site. This view is supported by the fractional value of the ratios Cl,/Alf observed which could be hardly explained by another mechanism.

Just as for a-TiC13, a very rapid extraction of chlorine takes place with p-TiC13 in the first moments of the reaction, and this fast exchange reaction affects again one chlorine atom for two titanium atoms close to the surface (Clv/TiBurf. ratio = 0.50).

The explanation given for the cu-TiCl3 systems also applied here since every chlorine atom in the surface may be considered as belonging to two titanium atoms.

>

V. REACTION BETWEEN a-TiCla AND AI(CH,)&I AT 20 AND 65°C.

The results of two experiments are summarized in Table 111. An important characteristic of these reactions is that the Cl/Al ratios are

exactly the same in the Al(CH&CI reactant and the volatile aluminum compound recovered after the reactions, even when the amount of A1(CH3)*- C1 is small in comparison to the chlorine present on the a-TiCl3 surface. However, the reaction may not be described as a chemisorption. In fact, methane and higher hydrocarbons are formed in comparatively high quanti- ties, and the infrared spectra of @-Ticla + A1(CH3),CI systems show the


TABLE I11 Reaction between a-TiCb and Al(CH&Cl, Method A

RB6 RB17

Step 1 Step 1 Step 2 Step 3 _ _ ~ ~

Step temperature, "C. Surface WTiCb, m.s Tisurf. Partial reaction time, sec. Total reaction time, sec. Al(CH&Cl fixed, mg. atom ZAl( CH3)&1 fixed, mg.-atom Al( CH&CI fixed/Ti.,,t. ZAl(CH&CI fixed/Ti.,,r. (CR + CD), m o l e

C E Cn(n > 1)

20 210 2.06 1800 1800 0.78 0.78 0.37 0.37

0.078 0.104

65 65 65 105 1.02 1800 3600 5400 1800 5400 10800 0.33 0.06 0.08 0.33 0.39 0.47 0.32 0.06 0.08 0.32 0.38 0.46

presence of A1CH3Cl2.* Moreover, the volatile products of the reaction contain small amounts of TiC13CH3, especially when the reaction is carried out at low temperature and stopped quickly. This compound is easily recognized because of its strong color. It is known that methane, ethane, and higher hydrocarbons are formed in considerable amounts during the decomposition of pure TiCl3CH8 in solutions and also during the decomposition of a mixture of TiC13CH3 and trimethyl- or chlorodimethylaluminum10 between 20 and 65°C. These facts lead one to assume that the first step in the reaction is a rapid exchange of C1 for CH,.

The TiC13CH3 formed partially decomposes, giving methane and higher hydrocarbons. The other product of the reaction, AlCHaC12, would be strongly associated with TIC& and might not be recovered by evaporation under vacuum. Such an effect is highly probable since Ziegler and co- workers have described several strongly associated products obtained from inorganic salts and chlorine containing aluminum compounds. 11.12

The proposed reaction scheme is as given in eqs. (1)-(4) : 2TiC13,,,tw + Al( CH3)zCl -P Ti2C16CH3 + A1CH3C12 (1)

Ti~C16cHs + TiClz-TiCI&H3 (2) (3)

TiCla + AICH3C12 + nonvolatile complex (4)

The formation of TiChCH3 tends to support the internal redox reaction:l3

TiClsCHa -P decomposition products (CHa + higher hydrocarbons + TiCl3)

2Ti+3=Ti+z + Ti+'

0 1 1 the basis of which we gave already a tentative explanation for the C1,/ T h . ratios of 0.5 observed in the fast exchange reaction between Tic& and

That the formation of TiCl3CH3 could be due to TiC14, present as an im- purity, is hard to believe since the values of the ratios Clv/Tisupf. of the rapid first exchange reaction are reproducible.

Al(CH3)S.l


Comparison of the value of (CR + CD), taking into account the presence of hydrocarbons higher than methane, and the amount of A1(CH3)2Cl used shows that reaction (3) is incomplete and leaves on the surface some Ti- CH3 bonds.

VI. REACTION BETWEEN a-TiCh AND AI(C2H&

The reaction was followed by means of method A, however slightly modified because of the low vapor pressure of A1(C2H6)3 as compared with that of Al(CH&. In order to recover the volatile products of the reaction it appeared necessary to increase the temperature; 85°C. was found satis- factory. The most important results are summarized in Table IV.

TABLE IV Reaction between a-TiCla and Al(CsH6)a at Various Temperatures, Method A

Reaction Reaction Reaction Reaction 112 106 105 103

Reaction temp., "C. Reaction time, min. Recovery temp., "C. Recovery time, min. CL/& (Ci - Cv)/Alr (CR + CD)/(ci - c v )

CdAlr Volatile phase

Cl/A1 CzHs/A1

30 11200 85 120 3.12 4.86 0.50 0.90

0.75 2.26

35 5590 85 180 2.05 4.30 0.50 0.81

0.98 2.02

65 60 85 180 1.00 4.09 0.41 1.17

1.00 2.50

65 135 85 -120 0.90 3.22 0.37 0.77

0.87 2.32

The symbols appearing in Table IV are defined as follows: Ci denotes the number of millimoles of C2Hs in the reactant A1(C2H6)3; C, is the number of millimoles of C2H6 in the volatile aluminum compounds recovered after the reaction; CR + CD is the number of millimoles of hydrocarbon formed during the reaction and recovery period; CH is the number of millimoles of hydrocarbon formed on hydrolysis of the nonvolatile reaction product.

Unfortunately, the surface of the a-TiCl3 samples was not measured. Neither were the hydrocarbon mixtures CR, CD, and CH quantitatively analyzed. It was proved that no ethylene was present in the gases CR and CD, which were essentially ethane. However, the presence of traces of polyethylene in the nonvolatile phase after hydrolysis points to the formation of ethylene in the reaction. Ci and C, were recovered as pure ethane on hydrolysis of the aluminum compounds. It may be concluded from Table IV that the aluminum alkyls present in the volatile phase after reaction consist of practically pure A1C1(CzH6)2, all of the initially added Al- (C2H6)3 having reacted. That no A1ClzCzH6 is recovered does not mean that this compound is not formed. It may be absent for various reasons: too low vapor pressure to be recovered by pumping, strong adsorption on


the Tic& surface, or disproportionation into AlClEtz and AlCl3 at 85°C. in vacuum.

The formation of Alcl(CzH& and the values of the Cl,/Alf ratios much greater than one suggest strongly that a mechanism similar to that proposed' for Al(CH& is operating.

a 2TiC13.,,f. + Al( C H Z H ~ ) ~ - Ti2C16C2H5 + l / 2 [AlCl(C2H&12

I

I t b 4. ( 5 )

l/z[Al( C~H6)3]2 decomposition products

It is highly probable that product I is unstable between 20 and 85°C. and that the decomposition step c of eq. (5) follows the mechanism proposed by de Vries for TiC13C2H6:9

TiC13+CH2-CH8

+ 2 Tic& + C2He + CzH4 (6) TiCI3~CH2-CH3 i5

Step c of reaction (5) seems to be important at low temperature, but at 65°C. the fixation of aluminum appears to be predominant. Indeed the ratio Cl,/Alf decreases from 3.12 to about 1 when the reaction temperature increases from 20 to 65°C.

Ti2C15C2H5

Ti~ClAl(CzHd4 I1

Therefore the following reaction must occur:

dJ. + Al(C2Ha)a 1/z[.Al(C2H6)312 (7)

Product I1 decomposes, giving ethane for nearly half of the ethyl groups involved in the reaction [(CR + C,)/(Ci-C,) N 0.51.

The difference in stoichiometry between the a-TiCl$-Al(CH& and the a-TiC13-A1(C2H& systems may be easily explained on the basis of the decomposition rate of the Ti-C bond, which is much greater for Ti-c&&, than for Ti-CH3.

and a reaction similar to step d of eq. (7) takes place for practically all the methyltitanium sites previously formed. It follows that the ratios Cl,/Alf and (Ci-C,)/Alf for reactions of long duration will be always close to 1 and 4.2 With Al- (C2&)3 on the contrary, the relative rates of steps c and d will be variable and will depend strongly on the reaction conditions (concentration, temperature, etc.) giving different Cl,/Alf and (Ci-Cv)/Alf ratios, as shown in Table IV.

With A1(CH3)3 the alkylated sites formed in reaction a react almost completely according to reaction (7) so that the sites become covered. A further reaction on the titanium chloride substrate then becomes impossible and the reaction is stopped, giving AlCl(CH& and leaving an excess of A1(CH3)3. On the

In the a-TiCl3-A1(CH& system step c of eq. (5) is

Another consequence of this situation is the following.


contrary, with A1(C2H6), the decomposition reaction of the Ti-C2H6 bonds formed leaves a clean titanium chloride surface which undergoes further alkylation.

With a given quantity of TiCL more A1(C2H6)3 will react than Al(CH&, and the reduction of titanium will also be more extensive in the former case. l4

Also, the composition of the surface compound will depend more strongly on the conditions of preparation. For instance, at low Al/Ti ratios in the reactant mixture, the conversion of A1(C2H5)3 may be complete and the clean titanium chloride surface regenerated by the decomposition of the unstable Ti-C bonds may now be attacked by AlCl(C&)2 formed in the alkylation reaction.

Finally it is worthwhile to point out that the decomposition of the Ti- CzH6 bonds results in the formation of chlorine vacancies on the surface. It has been shownl6Vl6 and it has been po~tulated"-~~ that the accessibility of titanium atoms, brought about by such vacancies, is a necessary condition for obtaining an active polymerization site.

One may therefore assume that those alkylating reactants which form unstable Ti-C bonds produce active sites of different kinds, and therefore, of different activities and different stereospecificity. The results and conclusions exposed above are difficult to reconcile with the mechanism proposed by Eden and Feilchenfeld.21 According to these authors, the first step in the reaction between TiCla and AI(C2HS)3 may be represented as shown in eq. (8).

c1 C2H6 CZH6 ,' '\\ / \

CZH6 I

TiCl, + A1--CzHs + TiCl:-, A1 . . . CzH6 -P TiCl,-l .Cl-Al + CzHs

AH6

Now product I11 is supposed to undergo an internal rearrangement and the final catalytically active compound would be IV :

c1 C4He / '\ / 'c1

C1,-2Ti

IV

This mechanism is unable to explain our stoichiometric results, especially the ratios Cl,/Alr and the ratios (Ci-C,)/AIf found. If the Eden and Feilchenfeld mechanism were valid, the ratios Cl,/Alf would be equal to 0 and the ratios (Ci-Cv)/Alf would be equal to 3. We found for similar conditions (room temperature and high Al/Ti ratio), respectively, 3.12 and 4.68 at 30°C. and 2.05 and 4.30 at 35°C.

Our results show also that the alkylating reaction takes place even at low temperature and seems to be very rapid. Therefore the concept of adsorption of the alkylaluminum on the Tic& surface is inadequate to describe the actual ~ i t u a t i o n . ~ ~ . ~ ~


M. CONCLUSIONS

The reactions between a- or B-TiCh and trialkylaluminum such as Al- (CH3)3 and A1(C2H5)3 seem to obey the following general mechanism: ( I ) a rapid exchange reaction in which one chlorine atom for every two titanium atoms in the surface of Tic13 is replaced by an alkyl group from (2) the titaniumalkyl compound either decomposes or adds a trialkylaluminum molecule and gives a product of general formula TiZCl5A1R4; (3) this product is unstable and decomposes, giving hydrocarbons and a nonvolatile surface product.

The comparison between the experiments with A1(CH3)3 and those with A1(C2H& shows that the decomposition of the Ti-C bond is relatively more rapid to the addition of aluminumalkyl with Al(CZH& than with A1(CH3)3.

Finally, this addition reaction is relatively more rapid to the exchange reaction with j3-TiC13 than with a-TiC13.

In the case of Al(CH&Cl the results obtained indicate that, after a first exchange reaction, the dichloromethylaluminum formed is adsorbed on the TiCL surface shielding it for further reaction.

The authors are grateful to Dr. B. Hargitay for stimulating discussions, to Mr. C. Leuk for his contribution to the experimental part, and to Mr. L. Bultot and Mr. V. Urbain for technical assistance.

References

1. Rodriguez, L. A. M., H. M. van Looy, and J. A. Gabant, J . Polymer Sci. A-1, 4

2. Rodriguez, L., and J. Gabant, J . Polymer Sci., 57,881 (1962). 3. Gutmann, V., H. Nowotny, and G. Ofner, 2. Anorg. Allgem. Chem., 278,78 (1955). 4. Korotkov, A. A., and T.-C. Li, V y s o k m l . Soedin., 3,686 (1961). 5. Natta, G., P. Corradini, I. W. Bassi, and L. Porri, Atti A d . Nazl. Lincei, Rend

6. Natta, G., P. Corradini, and G. Allegra, J . Polymer Sci., 51,399 (1961). 7. Cras, J. A., Nature, 194,678 (1962). 8. van Looy, H. M., L. A. M. Rodriguez, and J. A. Gabant, J . Polymer Sci. A-1, 4,

9. de Vries, H., Rec. Trav. Chim., 80,866 (1961).

1905 (1966).

Classe Sci. Fis. Mat 24, 121 (1958).

1927 (1966).

10. Rodriguez, L. A. M., and J. A. Gabant, unpublished results. 11. Ziegler, K., R. Koster, H. Lehmkuhl, and K. Reinert, Ann., 629.33 (1960). 12. Koster, R., and W. R. Kroll, Ann., 629,50 (1960). 13. Klemm, W., and E. Krose, 2. Anorg. C h a . , 253,209 (1947). 14. Cooper, M. L., and J. B. Rose, J . Chem. SOC., 1959, 795. 15. Hargitay, B., L. Rodriguez, and M. Miotto, J . Polymer Sci., 35, 559 (19.59). 16. Rodriguez, L. A. M., and J. A. Gabant, J. Polymer Sci., C4, 125 (1963). 17. Cossee, P., Tetrahedron Letters, 17,12 (1960); J . Caful., 3,80 (1964). 18. Adman, E. J., J . Polymer Sci., 62,530 (1962); J . Catal., 3,89 (1964). 19. Boor, J., Jr., J . Polymer Sci., C l , 257 (1963). 20. Adman, E. J., and P. Cossee, J . Caful., 3,99 (1964). 21. Eden, C., and H. Feilchenfeld, Tetrahedron, 18, 233 (1962). 22. Natta, G., and I. Pasquon, Advan. Catalysis, 11, 51 (1959). 23. Vesely, K.. J. Ambroz, and 0. Hamrik, J . Polymer Sci., C4, 11 (19S3).

1926 L. A. M. RODRIGUEZ, 11. M. VAN LOOY, AND J. A. GABANT

RQsumQ Les reactions suivantes ont 6tB Btudiees en absence de solvant: a-TiC1, + Al(CH&

A 2OoC, ,9-TiC13 + Al(CH& B 65OC, a-TiC1, + AICl(CH& ii 20 et 65OC et a-TiC13 + Al(CzH6)3 entre 30 et 65OC. Les reactions entre TIC13 et trialkylaluminum peuvent &re representees par un schema unique, les divergences s'expliquant par les differences de stabilite des liaisons Ti-CHa et Ti-CZH6. Ce schema est celui d6jA propose pour la reaction a-TiCl3 + Al(CH3)3 A 65°C. Dans le cas du systkme a-TiC13 + AlCl(CH3)Z il est montre que la reaction consiste en une alkylation de la surface de TiCl, probable- ment avec fixation du AlClZCH, forme.

Zusammenfassung Folgende Reaktionen wurden in Abwesenheit von Losungsmittel untersucht a-TiC13

+ Al(CH& bei 2OoC, @-Tic& + Al(CH43 bei 65"C, a-TiC4 + A(CH3)Z C1 bei 20 und 65OC. und a-TiC13 + !d(CZH& zwischen 30 und 65OC. Fur diese Reaktionen gilt ein allgemeines Reaktionsschema das in der vorgehender Arbeit ausfuhrlich entwickelt wurde. Unterschiede in der Gesammtstoechiometrie zwischen den verschiedenen Systemen werden auf Unterschiede der Stabilitat der intermediaren Ti-C Bindungen zuruckgefuhrt. Im Falle der Reaktion a-TiC1, + Al(CH&Cl muss wahrscheinlich eine Fixierung von AlCH3Clz im festen Reaktionsprodukt, in Betrachtung genommen werden.

Received June 24, 1964 Prod. No. 4819A